A county-level analysis of association between social vulnerability and COVID-19 cases in Khuzestan Province, Iran

Abstract

Social vulnerability is related to the differential abilities of socio-economic groups to withstand and respond to the adverse impacts of hazards and stressors. COVID-19, as a human risk, is influenced by and contributes to social vulnerability. The purpose of this study was to examine the association between social vulnerability and the prevalence of COVID-19 infection in the counties of Khuzestan province, Iran. To determine the social vulnerability of the counties in the Khuzestan province, decision-making techniques and geographic information systems were employed. Also, the Pearson correlation was used to examine the relationship between the two variables. The findings indicate that Ahvaz county and the province's northeastern counties have the highest levels of social vulnerability. There was no significant link between the social vulnerability index of the counties and the rate of COVID-19 cases (per 1000 persons). We argue that all counties in the province should implement and pursue COVID-19 control programs and policies. This is particularly essential for counties with greater rates of social vulnerability and COVID-19 cases.

1. Introduction

The COVID-19 pandemic has triggered a global public health disaster [1]. According to the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU), over 648 million cases of the disease have been reported worldwide, as of December 10, 2022. High incidence and mortality rates, unknown nature of the disease, lack of a definitive treatment solution, travel and social distancing restrictions, financial losses, fear of COVID-19 transmission, and increased stress were some of the major issues that faced of human society after the spread of COVID-19, especially in vulnerable groups [2].

The World Health Organization (WHO) proclaimed COVID-19 a pandemic on March 11, 2020, after evaluating the global case count and issuing early regulatory directives to execute healthcare measures against the new disease. Social distancing, mask use, and disinfectant use were all promoted as the main measures for reducing the occurrence of COVID-19 [3]; [4]. According to research, disasters, economic shocks, and other disturbances affect various segments of society in markedly different ways [5]; [6]. Indeed, pandemics and other natural and man-made hazards disproportionately harm the most vulnerable persons and communities [7] [8,9]. The Centers for Disease Control and Prevention (CDC) uses ‘social vulnerability’ to determine which communities are more vulnerable to adverse events such as disasters triggered by natural hazards or disease outbreaks [10]. The degree to which a society can prepare for and respond to a natural or man-made hazards, such as a hurricanes, chemical spills, or disease outbreaks, is defined as social vulnerability [11]. Social vulnerability is determined by the social, economic, demographic, and geographic aspects that influence risk exposure, as well as the capacity of society to cope with risk [12]. The socio-economic conditions, disability, population composition (by age, race, and ethnic origin), housing status, family structure, social security, and public health situation are all considered indicators of social vulnerability [13]. Social vulnerability has a substantial impact on mortality and health outcomes as well [14]. It is anchored in social systems that result in unequal risk exposure and social problems [15].

Over the last decade, research efforts have shifted toward socio-environmental vulnerabilities as a result of their relationship with geophysical and human risks. With the emergence of the COVID-19 pandemic, there has been renewed attention to the association between social vulnerability and pandemics [16]; [8,[17], [18], [19], [20], [21]]. Additionally, researchers have attempted to develop COVID-19-specific vulnerability indicators. Since the outbreak began, the number of studies demonstrating the detrimental impacts of COVID-19 on socially vulnerable individuals has expanded tremendously [22]. Several early studies examined the relationship between social vulnerability and COVID-19 mortality in the United States. Khazanchi et al. [19]; using a Poisson-like regression approach and examining the cities that feature different socio-economic groups, types of housing, and transportation infrastructure distribution, discovered that those who live in counties with the highest social vulnerability face an increased risk of infection and mortality [19]. In another study, Nayak used a linear model to assess 433 US counties and discovered that high levels of social vulnerability were associated with greater COVID-19 mortality [23]. Additionally, Kim and Bostwick found a positive association between the percentage of black residents in the Chicago Census tracts and social vulnerability. Further, they discovered that spatial clusters of social vulnerability are connected with an increased risk of death from COVID-19. Moreover, other studies have revealed a strong correlation between social vulnerability and COVID-19 mortality in the United States [11]. Also, there is evidence that COVID-19 has a greater impact on ethnic and racial minorities, the elderly, and those who are socially and economically disadvantaged [24]. Numerous studies have established that African Americans have a greater rate of incidence and mortality than non-Hispanic whites [25,26][27].

Existing literature clearly shows that the COVID-19 pandemic has resulted in an increase in inequality in different societies, especially among vulnerable groups [[28], [29], [30]]. Moreover, one of the tangible effects of the COVID-19 pandemic, as mentioned, has been the changes in the economic status of urban communities, especially in developing countries. These changesdirectly affect the workers and suppliers of the private sector, who are among the most vulnerable groups. Indeed, social distancing and/or quarantine measures to contain COVID-19 have disrupted various economic activities. The pandemic and measures developed to control it have disproportionately impacted some socio-economic groups with limited planning, coping, and recovery capacities. These include, for instance, minorities and those employed in the informal sector [[31], [32], [33]].

Existing literature demonstrates that socially vulnerable people are more prone to disasters and that social disparities foster illness transmission, complicating efforts to manage the pandemic. Indeed, existing social disparities increase vulnerable and marginalized populations' risk of illness and mortality from COVID-19 [34]. The social sphere and social vulnerability are critical components of COVID-19 research. Accordingly, the study of social factors is critical for responding to this risk. Empirical evidence related to social vulnerabilities in the context of the pandemic has been reported for different countries, including the United States and developing countries such as India [35,36], Brazil [7,37], Philippines [38], and Palestine [39]. However, no study has been conducted in Iran or Khuzestan region to examine the association between social vulnerability and COVID-19 cases or deaths.

Iran was one of the first epicenters of the COVID-19 in the Middle East, and, like many other countries, it has gone through multiple waves of the pandemic [40,41]. Khuzestan province, located in southwestern Iran, has one of the greatest concentrations of COVID-19 [42]; [43]; [44]. Like other provinces in Iran and developing countries, Khuzestan faces the issue of social inequality. Lack of infrastructure in villages and distant locations, high unemployment in suburban areas, and high prevalence of poverty in cities and villages are some of the most critical issues. These issues make it challenging to deal with natural and man-made hazards. . The Province, as one of the most populous provinces of Iran, is located in a multi-hazard area and suffers much damage due to various hazards such as floods, earthquakes, landslides, and dust. These multiple hazards have affected the socioeconomic conditions of residents in cities and villages [45]. The province is exposed to numerous natural hazards and is one of the most deprived provinces of Iram. It has a high unemployment rate, and the occurrence of adverse events (e.g., the COVID-19 pandemic) has worsened its unfavorable conditions, especially for deprived groups. Assessing the province's social conditions and addressing vulnerabilities are essential for developing policies and programs to battle the crisis and increase the province's overall resilience. Therefore, this study aimed to examine the association between social vulnerability and the COVID-19 cases in the province by developing a methodology based on Multi-criteria Decision Making (MCDM) and geographic information systems. In other words, we wanted to examine in there is an association between social vulnerability and the prevalence of COVID-19 in the province. We investigate the vulnerability of the counties in Khuzestan Province and explore the relationship between the composite index of social vulnerability and the COVID-19 incidence rate. Our findings could provide planners with a clearer view of the impacts of social vulnerability on the COVID-19 in the field of planning and implementation.

2. Theoretical framework

2.1. Social vulnerability

The concept of vulnerability has been used since the 1970s, especially in studies of crisis management, development, and the environment [46]. Vulnerability is the degree to which a system is prone to damage and cannot adapt to the harmful effects of a change. Different viewpoints on social-ecological system vulnerability show that vulnerability is affected by the expansion or weakening of the elastic characteristics of social and ecological elements, affecting the system's ability to adapt to incoming shocks [47]. Vulnerability is divided into internal vulnerability (ability to cope with new conditions) and external vulnerability (linked to being exposed to risks). External vulnerability mainly refers to the structural dimensions of vulnerability and risk, and the concept of internal vulnerability is associated with the measures needed to deal with economic-social and environmental changes. External vulnerability is often more highlighted, while internal vulnerability is difficult to identify and measure [48]. Thus, it can be argued that vulnerability refers to conditions that, due to physical, social, economic, and environmental factors, determine and increase the ability of societies against damage caused by hazards [49][50].

Social factors play an important role in reducing or increasing human vulnerability. The issue of social vulnerability to hazards received more attention from researchers in the 1970s, when they realized that vulnerability can include, in addition to the physical dimension, social and economic factors that affect the resilience of society [51]. Accordingly, social vulnerability refers to socio-demographic factors that affect society's ability to respond to and recover from stressful factors at the community level, such as epidemics and natural disasters [52]. Social vulnerability refers to a condition when certain individuals or communities have limited ability to mitigate, absorb, recover from, and adapt to threats and disasters. It is a combination of factors that determine people's living standards and includes livelihood, wealth, and other assets that are endangered by identifiable events in nature and society [53]. The increased social vulnerability could undermine community resilience [54][55]. Studies conducted around the world have shown that different groups of people tend to show varying degrees of vulnerability to hazards depending on their standard of living and social and economic situation in different parts of the world. Assessing the social vulnerability of human settlements against hazards, as one of the basic indicators in risk assessment, has a special place in the crisis management cycle, and without an awareness of the social and economic situation of the residents, it is not possible to assess their vulnerability to hazards. Different techniques including statistical analysis and geographic information systems (GIS) have been used to measure and evaluate social vulnerability indicators. For instance, Ebert et al. [56] evaluated social vulnerability using ground sampling and spatial measurements through satellite images and GIS data. Moreover, Armaș et al. (2013) used a multi-criteria analysis model to assess the social vulnerability of Bucharest [57]. Furthermore, principal component analysis (PCA) and geographic information systems (GIS) were used to measure social vulnerability to floods [58], earthquakes [59], and COVID-19 pandemic [35,60]. The analytic hierarchy process (AHP) and geographic information systems were also used to investigate social vulnerability against COVID-19 [39] and local Spearman's rank correlation coefficient was employed to measure the relationship between social vulnerability and COVID-19 [34]. A review of the methods used in related studies shows that the social vulnerability index was not calculated as a composite index. Besides, weighting techniques have not been used due to the different impacts of the indicators.

2.2. Social vulnerability and COVID-19

The rapid spread of the new coronavirus (COVID-19) has brought devastating impacts and damages to countries around the world and created new challenges to achieving sustainability [61]. Initially, COVID-19 was misleadingly named the ‘great equalizer’, meaning that everyone is equally vulnerable to the virus, and people's economic activity, infection rates, deaths, etc. Are similarly affected regardless of their social status [62]. However, empirical evidence has shown that although the impact of COVID-19 has been widespread, not all regions or social groups have been equally affected by this epidemic [63,64]. COVID-19 widely affects socioeconomic activities, work life, and food systems [65]. The analysis of infection and mortality rates has shown that specific individual and spatial characteristics, income levels, and social positions of people, especially multidimensional vulnerability criteria are associated with a higher or lower probability of infection with the COVID-19 virus [66][67]. Accordingly, the COVID-19 pandemic has had major impacts on life everywhere, but some countries and communities have been affected disproportionately [68]. When disasters of any kind occur, socially vulnerable people are at the greatest risk. People who live in vulnerable situations include people facing systematic deprivation and those who have been discriminated against depending on socio-demographiccharactersitics such as age, gender, faith, physical ability, ethnicity, and income. . Moreover, people who live in inappropriate housing, are exposed to unfavorable environmental conditions, and are at climatic risk are more socially vulnerable [69].

Reports of morbidity and mortality from COVID-19 show higher rates among racial/ethnic minorities, older adults, low-income groups, and less educated people [70]. For example, crowded and poor districts, where minorities are overrepresented, render social distancing orders designed to control the spread of the virus less practical and increase the risk of infection. Vulnerable groups inadvertently reside in unfavorable and overcrowded neighborhoods. As a result, they are often more exposed to health risks and have limited ability to comply with health and sanitation measures [30]. Furthermore, due to their unique economic conditions, vulnerable communities cannot afford to stay home and respect social distancing measures [64]; [71].

Therefore, the COVID-19 epidemic has not only caused a public health emergency but has also led to major economic and social crises, with unequal distribution of its consequences throughout economies and societies [72]. Moreover, studies have confirmed that social inequalities during pandemics cause the risk of unequal distribution of the disease among the population. Furthermore, the unavailability of resources and difficulties in access to basic health or preventive information increase social inequalities during pandemics [29,30]. As a result, socially vulnerable people experience higher exposure to health risks and are more likely to be negatively impacted [73]. People who live in vulnerable conditions are exposed to significantly exacerbated inequality gaps caused by the pandemic and are more likely to suffer from possible negative and long-term physical, socio-economic, and psychological health consequences [28].

3. materials and methods

This is a multi-criteria study that utilized MCDM and correlation methods. SPSS and ArcGIS software programs were employed to analyze the data. In this section, we first briefly introduce the study area and then explain the data and analysis methods.

3-1. study area

Khuzestan province, with an area of 632,528 km², is located in southwestern Iran and northwest of the Persian Gulf (Fig. 1 ) [74]. The province's altitude ranges from 0 to 3740 m. Also, the climate varies from cold in the north to warm in the south [75]. According to the Statistical Center of Iran (2016), the province consists of 27 counties and 54 cities, and the total population is 4,700,000 [42]. It is one of the provinces that has been greatly affected by COVID-19 due to its business relations with other countries and daily work migrations across this province. Like many provinces of Iran, it has left behind five COVID-19 peaks. Due to the absence of effective policies and planning measures, the province is exposed to many hazards, especially dust, environmental changes, and climate change. Some parts of the province are more deprived, and there are economic and social inequalities between cities and villages and between different urban areas [76]. These inequalities, at the time of adverse events, result in varying degrees of vulnerability and resilience among different groups.

Fig. 1.

Location of khuzestan province.

According to the statistics of 2021, the marginal population of Khuzestan Province is 1,747,739 persons, accounting for 37% of the province's population. As mentioned, the marginal population lives in the cities, especially Ahvaz. A large part of the population living in marginal areas are people who migrated to cities due to social and spatial inequalities [77].

3.2. Data

This research relies on secondary data and primary data obtained via expert surveys.

The dataset in this study included social vulnerability data and COVID-19 data. The social vulnerability data were collected from the 2016 Census of Khuzestan Province, the Statistical Yearbook of Khuzestan Province (2017), and the economic, social and cultural report of Khuzestan Province (2016–2017). As this study aimed to assess vulnerability at the county scale, the indicators were measured for different counties across the province. The data collected for the counties were used as the raw matrix in the decision-making model. In the first stage, the indicators with different (incremental and decremental) scales and different intervals were standardized.

The COVID-19 data were recorded separately for cities and villages by Ahvaz Jundishapur University of Medical Sciences, which is theCOVID-19 management center in the province. The data for the cities and villages were normalized in the form of the COVID-19 infection rate per 1000 people for each county. Considering the number of COVID-19 peaks and given the data accessibility issues, the statistics of COVID-19 patients were prepared for a year from October 2020 to September 2021. We used a longitudinal approacc [1],.

3.3. Research indicators

Indicators are statistical tools applied to describe complicated social concepts in scientific analysis. Scientific and accurate indicators that account for a wide range of individual and group differences are the foundation of social vulnerability research. Due to the complexity and dynamic nature of the social vulnerability, there are multiple indicators in this scope. Therefore researchers have employed different indicators based on the needs of their projects. The crucial point is to identify those key indicators that accurately indicate changes.

Context and local conditions are critical in the selection process of social vulnerability indicators. In the United States, one of the indicators is the percentage of foreign-language speakers [10]. Muslim communities have been identified as vulnerable groups in developing countries such as India [35], whereas in Iran, households supported by Relief Foundationsappear to be an indicator of household living conditions; relief foundations are institutions set up in some countries to identify and assist vulnerable families. Therefore, the families and individuals whom these institutions support do not have the appropriate socio-economic conditions to deal with risks [78].

Fatemi et al. [79] have identified the most important social vulnerability indicators in Iran, including gender, demographic features, socio-economic conditions, disability and special needs, and public resource availability. They stated that these indicators provide an accurate assessment of society's vulnerability to technological hazards and man-made disasters, particularly in developing countries such as Iran.

Based on the existing literature and considering the context-specific conditions, the indicators listed in Table 1 were used in this study. A brief description of each indicator is provided in the remainder of this section.

Table 1.

Indicators and variables of social vulnerability.

Indicator	Variable	Effect
Age	Rate of the elderly over 65 years	Increasing
Economic	Rate of population dependency	Increasing
	The per capita income	Decreasing
	The unemployment rate	Increasing
Education	Total literacy rate	Decreasing
	Female literacy rate	Decreasing
Population	Population density index	Increasing
	Female population	Increasing
	Households dimension	Increasing
	Female-headed households	Increasing
	Rural population	Increasing
	Rural households	Increasing
Social Security	Households supported by the Relief Committee	Increasing
	Rate of pension households	Increasing
Residence	Number of inhabited villages	Increasing
	Rate of rural households with safe water	Decreasing
	Number of villages per 100 square kilometers	Increasing
	Rate of villages under 20 households	Increasing
	Rate of residential units with two households and more	Increasing

Income and poverty are critical drivers in the field of social vulnerability. Income can affect other indicators such as education, job type, overcrowding, vehicle and homeownership, and unemployment. Additionally, low education levels contribute to poverty, overcrowding, unemployment, income inequality, and marginalization [80]. Women face greater limits and have less access to resources than men, especially in developing countries. These lead to more poverty and risk vulnerability [81]. Increase in female-headed households can also increase social vulnerability [5] [82]. Weak socio-economic conditions and the quality of housing units show the state of public health affected by disasters. Unemployment is associated with an increase in vulnerability during times of crisis and disaster [79]. Poverty and lack of access to resources affect the vulnerability of individuals and households. Unemployment exacerbates the difficulty of low-income households remaining quarantined, thereby increasing their vulnerability. Those with low socio-economic conditions may not be able to afford to buy disposable face masks. Moreover, they may also be unable to wash reusable cloth masks and rely on public laundry, which increases the exposure risk [1]. Education level is directly related to risk awareness and comprehension. Education plays a significant role in reducing vulnerability since it has an effect on people's awareness and understanding of disasters. Individuals with a low degree of education frequently face economic difficulties. Additionally, low levels of education decrease an individual's sense of control over health choices, which might be associated with poor health outcomes [13]. Houses that are unsafe and crowded make social distancing measures, health care, and proper access to water ineffective. Thus, crowded households with large families sharing rooms may be unsuitable for social isolation and social distancing, increasing the risk of COVID-19 transmission at home [83]. Social vulnerability is significantly related to demographic characteristics such as female population, age, and households with a disabled member. In addition, children under the age of 14 and people over the age of 65 (at opposite ends of the age range) have a relatively limited potential for self-protection during adverse events [84]. The elderly are also likely to be more vulnerable to adverse events. Due to their physical state and lack of adequate immunological responses, they face difficulties in coping with and responding to adverse events, including pandemics. Woolf et al. [85] noted that being 65 or older is a significant risk factor for COVID-19 mortality [85]. A high rate of dependency also leads to increased social vulnerability [38]. The dependency ratio is calculated based on the percentage of the population under the age of 16 and over the age of 65, along with people with physical disabilities, relative to the population between the ages of 16 and 65. Indeed, this is the ratio of unemployed to employed people, which reflects the pressures on the employed people in a society. Households with a significant number of dependents confront numerous difficulties due to their limited financial opportunities [86]. The households dimension index quantifies the impact of poor housing conditions on vulnerability [87]. This is because larger households will have a greater number of dependent members (economically inactive) [88].

Population density could be a risk factor by increasing risk exposure. When natural or human hazards occur, higher population densities could result in increased losses due to inefficient management and planning [89,90]. In developing country cities, population density is also considered an important factor in the outbreak of COVID-19, and avoiding situations with higher population densities is an essential requirement for limiting the prevalence of COVID-19 [91]. Moreover, an increase in the number of inhabited villages within a geographical region complicates service delivery and relief activities, which may result in increased vulnerability. The number of villages per 100 square kilometers frequently challenges the provision of effective products and services in rural areas, resulting in rural vulnerability and limited access to sanitation [35]. Rural populations in the Khuzestan province have insufficient access to public services, health care, and infrastructure. Furthermore, rural households face undesirable employment, income, and education conditions. Dintwa et al. [92] stated that rural areas in Botswana have a greater vulnerability ratio and a lower capacity for risk management than urban areas [92].

Table 1. Indicators and variables developed to assess the social vulnerability of the counties in Khuzestan province.

3.4. The analysis process

The overall process is shown in Fig. 2 . First, the sum of indicators for each county was calculated using the decision-making model as a variable in the correlation analysis. The analysis was performed in two stages. In the first stage we ranked the 27 counties in Khuzestan province based on the 19 indicators listed in Table 1. In the second stage, the final result of the ranking model is used as the first variable, and the rate of COVID-19 cases per 1000 persons is used as the second variable. Indeed, counties and indicators were included as options and criteria in the raw matrix, respectively, and then prioritization and ranking were performed. We employed the complex proportional assessment (COPRAS) method, which is further discussed later, to rank the counties. In addition, the decision-making trial and evaluation laboratory (DEMATEL) method was used to determine the indicators' weights. We asked ten subject-matter experts to complete the DEMATEL questionnaire. Employing non-experts in decision-making processes creates complications and decreases efficiency. Therefore, the group of decision-makers included managers and researchers with expertise in geography, urban planning, and crisis management. They have been working on the subject for over a decade and are well aware of effective variables and indicators of social vulnerability. We purposefully selected specialists through face-to-face meetings and in-depth discussions with relevant individuals. It should be noted that during the three months, five questionnaires were physically distributed in the area, and five questionnaires were sent online to related researchers.

Fig. 2.

Flowchart of methods used for the purpose of this research.

In the second stage, the Pearson correlation was used to determine the relationship between the integrated social vulnerability index and the rate of COVID-19 cases. The Pearson correlation has been previously utilized to evaluate the relationship between social vulnerability and COVID-19 [35]. The relationship between social vulnerability and the rate of COVID-19 cases in this study is correlational, indicating whether the rate of COVID-19 cases is higher in the counties with a high level of vulnerability. The Pearson correlation coefficient is used to determine the correlation between two variables when both are on a ratio scale and interval scale. This study's data are on a ratio scale.

According to Alkan and Kahraman [93]; MCDM procedures efficiently resolve complicated situations using a variety of criteria. Due to the nature of the social vulnerability, using multi-criteria decision making is a recurring method [[94], [95], [96], [97], [98]] [99]. Multi-criteria decision making methods such as analytic hierarchy process (AHP), analytical network process (ANP), best-worst method (BWM), the technique for order of preference by similarity to ideal solution (TOPSIS), and vlse kriterijumsk optimizacija kompromisno resenje (VIKOR) have been exploited in COVID-19 studies [[100], [101], [102], [103], [104]][105]. For instance, Malakar [4] assessed social vulnerability to COVID-19 using MCDM methodologies (fuzzy AHP and fuzzy TOPSIS).

The use of mixed methods is a new approach in the MCDM and MCDA [106]. So, DEMATEL approach for weighting criteria was combined with COPRAS method for ranking choices as will be further discussed below [107,108] [109].

Geographically weighted regression (GWR) and exploratory factor analysis (EFA) have been utilized in related studies. GWR was not practicable due to the small number of units (county), and in EFA, the summary and classification of indicators encountered mistakes. The advantage of MCDM analysis is that it provides acceptable results even when the number of available options is limited. Moreover, variations in the relative importance of the indicators result in different outcomes.

DEMATEL Method.

DEMATEL method is implemented in steps as follows:

Step 1

Forming direct connection matrix

To form a direct connection matrix, experts' opinions are applied, and then, to form the total matrix, the opinions of all experts are taken into account. Therefore, a direct connection matrix is formed using the verbal concepts defined in Table 2 .

Step 2

Normalizing the direct connection matrix:

To normalize the matrix, (1), (2) are employed:

$${\mathrm{H}}_{\mathrm{i}\mathrm{j}}=\frac{{\mathrm{z}}_{\mathrm{i}\mathrm{j}}}{\mathrm{r}}$$ (1)

R is obtained from the following one:

$$\mathrm{r}={\max }_{1\le \mathrm{i}\le \mathrm{n}}$$ (2)

Step 3

Calculating the total connection matrix:

After calculating the above matrices, the fuzzy total connection matrix is obtained according to Formula (3):

$$\mathrm{T}=\underset{\mathrm{k}\to +\infty }{\lim }\left({\mathrm{H}}^1+{\mathrm{H}}^2+\dots +{\mathrm{H}}^{\mathrm{k}}\right)=\mathrm{H}\times {\left(\mathrm{I}-\mathrm{H}\right)}^{-1}$$ (3)

Where, I is the identity matrix in formula (3).

Step 4

Calculating the sum of the rows and columns of the matrix

The sum of rows and columns is obtained according to (4), (5):

$${\left(D\right)}_{n\times 1}={\left[\sum _{j=1}^n{T}_{ij}\right]}_{n\times 1}$$ (4)

$${\left(R\right)}_{1\times n}={\left[\sum _{i=1}^n{T}_{ij}\right]}_{1\times n}$$ (5)

Where, D and R are n * 1 and 1 * n matrices, respectively.

Step 5

Calculating the weights of influence and the effectiveness of the criteria

The relative importance of the criteria is calculated using formula (6):

$$w_j={\left[{\left(D_i+R_i\right)}^2+{\left(D_i-R_i\right)}^2\right]}^{\frac{1}{2}}$$ (6)

Step 6

Normalizing the weights of the criteria

The weights obtained from the previous step can be normalized using formula (7) (LIU & WU, 2004)

$${\bar{W}}_j=\frac{{\mathit{W}}_{\mathit{j}}}{\sum _{\mathit{j}=\mathbf{1}}^{\mathit{n}}{\mathit{W}}_{\mathit{j}}}$$ (7)

COPRAS Method.

The COPRAS method was first proposed in 1996 by Zavadskas and Kaklauskas at Vilnius Gediminas Technical University [110]. The COPRAS method is implemented in steps as follows:

Step 1

Forming a matrix

The COPRAS method begins by defining the weights for each criterion. The choice matrix is a two-dimensional matrix that contains both options and criteria. Additionally, the matrix contains a column indicating the weight assigned to each criterion. To complete the matrix values, we calculated the value of each criterion independently for each option and entered it in the appropriate location.

Step 2

Forming a collective decision matrix

This phase entails the aggregation of opinions, which can be accomplished using the arithmetic mean.

Step 3

Forming a weighted matrix

To weigh the decision-making matrix, multiply the values of each option by their weight and divide by their sum. The weighted decision-making matrix is constructed using the following formula:

$$d_{ij}=\frac{q_i}{\sum _{j=1}^n{x}_{ij}}{x}_{ij}$$ (8)

Where, qi is the weight assigned to each criterion, and xij is the value assigned to each option for each criterion.

Step 4

Calculating the value of positive and negative criteria

Then, the positive and negative criteria are discovered and separated. While increasing the value of a positive or consistent criterion raises its desirability, increasing the value of a negative criterion diminishes its desirability. After determining the positive and negative criteria, it is necessary to establish their final values. Formula (9) is used to calculate the indexes Sj + and Sj-for this purpose.

$$d_{ij}{S}_j^{-}=\sum _{z_i=-}{d}_{ij}{S}_j^{+}=\sum _{z_i=+}{d}_{ij}$$ (9)

Step 5

Calculating the final value of options

The algebraic sum of positive and negative values is calculated independently using Formula (10). Formula (10) is used to determine the final value of each option (Q) in the final step:

$$Q_j=S_j^{+}+\frac{s_{\mathit{min}}^{-}\sum _{j-1}^n{s}_j^{-}}{s_j^{-}\sum _{j-1}^n\frac{s_{\mathit{min}}^{-}}{s_{\mathit{min}}^{-}}}=s_j^{+}+\frac{{\sum }_{i=1}^n{s}_j^{-}}{s_j^{-}\sum _{i=1}^n\frac{1}{s_j^{-}}}$$ (10)

Sj + is the algebraic total of the positive requirements for each option in Formula 10, whereas Sj - is the algebraic sum of the negative criteria for each option. In this section, the number 1 is divided by Sj- and then the value of Q for each option is computed using the preceding formula. The Q value represents the relative importance and weight of each option in terms of the criteria. A high value indicates the significance and desirability of the majority of alternatives.

Table 2.

Quantitative values equivalent to the verbal concepts of the initial matrix.

Verbal Concepts	Quantitative Values
No Impact	0
Very little impact	1
Low impact	2
Strong Impact	3
Very strong impact	4

4. Results

As mentioned in the previous section, the weights were determined using DEMATEL method. The weights acquired are shown in Table 3 . The population density index is the most significantly weighted, followed by the households dimension, unemployment rate, female-headed households, and female population. These weights show the indicators' priority and significance to experts. As each indicator's weight increases, its impact on the county's social vulnerability increases accordingly.

Table 3.

Weights determined by DEMATEL Method.

Indicator	Weight
Population Density	0.0713
Households Dimension	0.0702
Unemployment Rate	0.0670
Female-headed Households	0.0655
Rural Households	0.0628
Rate of Pension Households	0.0621
Households Supported by the Relief Committee	0.0599
Rate of Residential Units with two Households and More	0.0584
Literacy Rate	0.0541
Rate of Population Dependency	0.0493
Rural Population Density	0.0489
Elderly over 65 Years	0.0486
Number of villages per 100 Square Kilometers	0.0469
Number of Inhabited Villages	0.0465
Rate of Rural Households with Safe Water	0.0426
Rate of villages under 20 Households	0.0386
The per capita income	0.0383
Female Literacy Rate	0.0374
Female Population	0.0319

The generated weights were used in the COPRAS model. The COPRAS model divides indicators into two categories: positive and negative. Positive signs contribute to an increase in vulnerability, whereas negative indicators contribute to a decrease in vulnerability. The total literacy rate, per capita income, the proportion of rural households with safe water, and the female literacy rate were exerted as negative indicators in this study, while 15 additional factors were utilized as positive indicators in the COPRAS model. The COPRAS model's final output is shown in Table 4 . The most vulnerable county is Ahvaz, followed by Indika, Dezful, Izeh, Baghmalek, Shadegan, Lali, Ramhormoz, Shushtar, Shush, Karun, Hamidiyeh, Dashte Azadegan, Behbahan, Abadan, Andimeshk, Bavi, Bandar Mahshahr, Ramshir, Masjed Soleiman, Gotvand, Hoveyzeh, Aghajari, Khorramshahr, Haftkol, Omidieh, and Hindijan.

Table 4.

Ranking of the Counties based on Social Vulnerability Indicators.

County	Q	Rank
Ahvaz	0.0584	1
Andika	0.0489	2
Dezful	0.0477	3
Izeh	0.0466	4
Baghmalek	0.0418	5
Shadegan	0.0416	6
Lali	0.0405	7
Ramhormoz	0.0404	8
Shushtar	0.0384	9
Shush	0.0383	10
Karun	0.0382	11
Hamidiyeh	0.0374	12
Dashte-Azadegan	0.0357	13
Behbahan	0.0355	14
Abadan	0.0346	15
Andimeshk	0.0346	16
Bavy	0.0345	17
Bandar-e-Mahshahr	0.0338	18
Ramshir	0.0337	19
Masjed-Soleiman	0.0320	20
Gotvand	0.0317	21
Hoveyzeh	0.0315	22
Aghajari	0.0310	23
Khorramshahr	0.0308	24
Haftkel	0.0292	25
Omidiyeh	0.0270	26
Hendijan	0.0263	27

According to the integrated vulnerability index, the counties of Khuzestan province are classified into five classes (Fig. 3 ). Ahvaz is the first-level county. Andika, Dezful, and Izeh are the second level. The counties of Baghmalek, Ramhormoz, Shadegan, and Lali are on the third level. Behbahan, Ramshir, Bandar-e-Mahshahr, Abadan, Karun, Bavy, Hamidiyeh, Shushtar, Dashte-Azadegan, Shush, and Andimeshk are located on the fourth level. Haftkel, Masjed-Soleiman, Gotvand, Hoveyzeh, Omidiyeh, Aghajari, Khorramshahr, and Hendijan comprise the last level.

Fig. 3.

Spatial display of social vulnerability in the counties of khuzestan province.

The COVID-19 data of Khuzestan counties were utilized to investigate the relationship between vulnerability and the rate of COVID-19 cases. The raw COVID-19 data and the rate of cases per 1000 people are shown in Table 5 .

Table 5.

COVID-19 data in counties of khuzestan province.

County	Rate of Covid-19 Cases	Ratio per 1000
Omidiyeh	7390	80
Andika	2553	54
Andimeshk	11,111	65
Ahvaz	78,312	60
Izeh	11,920	60
Abadan	24,442	82
Aghajari	2121	120
Baghmalek	7390	70
Bavy	5189	54
Bandar-e-Mahshahr	15,853	54
Behbahan	22,318	124
Hamidiyeh	2152	40
Khorramshahr	16,259	95
Dezful	29,096	66
Dashte-Azadegan	7862	73
Ramshir	3899	72
Ramhormoz	10,694	94
Shadegan	12,002	87
Shush	5956	29
Shushtar	23,198	121
Karun	4350	41
Gotvand	5178	79
Lali	2770	73
Msjed-Soleiman	11,536	102
Haftkel	3729	169
Hendijan	3668	95
Hoveyzeh	1547	40

According to the results of the Pearson correlation test (Table 6 ), there is no statistically significant correlation between the integrated social vulnerability index and the rate of COVID-19 cases, as the sig value exceeds 0.05. Fig. 4 shows the county-level social vulnerability and the rate of COVID-19 cases in Khuzestan province.

Table 6.

Pearson correlation between social vulnerability and the rate of COVID-19 cases.

		Covid-19 Cases Rate (per 1000 people)
Integrated index of social vulnerability	Pearson Correlation	−.355
	Sig. (2-tailed)	.069
	N	27

Fig. 4.

Distribution of social vulnerability and the rate of COVID-19 cases in the counties of Khuzestan province.

5. Discussion

Assessment and comprehension of the impact of COVID-19 on social vulnerability at the local and regional levels will help develop resilience programs [34]. Besides, addressing inequality is the first step in improving sustainability and risk management approaches during COVID-19 [73]. The purpose of this study was to determine social vulnerability in the counties of Khuzestan province and to investigate its relationship to the COVID-19 cases rate. To address the first research objective, the multi-criteria decision-making approach was used for ranking the counties and an integrated social vulnerability index was developed for each county According to the findings of the multi-criteria decision-making process, Ahvaz county (the province's capital) and counties in the province's east have the highest levels of social vulnerability. Additionally, the counties' conditions varies in terms of COVID-19 cases. To address the second objective of the research, the Pearson correlation was used The most notable finding is that no relationship exists between counties' level of social vulnerability and the rate of COVID-19 cases. This indicates that COVID-19 spread does not correlate with the rate of social vulnerability. Ahvaz county ranked first in terms of social vulnerability. Ahvaz County's vulnerability is exacerbated byits high population density and a high share of poor and marginalized neighborhoods. Indeed, while the county has a high proportion of urban residents, a significant proportion of them live in poverty. These residents have migrated from other villages and cities to the province. Furthermore, Ahvaz city, with 20 marginalized neighborhoods, has the most marginalized areas in Iran. These marginalized neighborhoods are characterized by the lack of water and energy infrastructure, inadequate sanitation, high unemployment rate, lack of educational services, and the lack of healthcare facilities. Besides, most of the residents in these neighborhoods are engaged in informal jobs, including people who work at home or on the street, daily wage workers, and service workers who work in other people's houses [31]. COVID-19 has directly and severely affected the employees and workers in this sector [111]. Omidiji et al. [112] stated that the COVID-19 virus increases the vulnerability of people from disadvantaged classes, especially in informal settlements and slums. Due to the significant rural population in the northeastern counties of Khuzestan province, the level of social vulnerability has increased since villagers in the province face worse conditions in terms of employment, access to services and facilities, and exposure to hazards (especially natural hazards). As a result of the large number of villages in some counties, such as Indika, Bagh-Malek, and Dezful, as well as suburbs in others, like Ahvaz, Mahshahr, and Abadan, there is no significant difference in social vulnerability index scores between the counties.

Our findings contradict Neelon et al. [1]; Khazanchi et al. [19]; and Karaye & Horney [10] in the United States, as well as Sarkar & Chouhan [35] in India. They discovered that counties that were more socially vulnerable had a higher rate of COVID-19 cases. Social vulnerability varies according to context. These contradictory results could be attributed to the varying socioeconomic conditions in different countries. For example, in the United States, language groups and minorities (blacks) have worse conditions and a higher rate of COVID-19 [18,[24], [113], [114]]. Whereas, in developing countries such as India, more low-income households live in harsher conditions than in Iran [35] [36]. In the case of Khuzestan province, it was found that all ethnic groups face similar living conditions and comparable levels of social vulnerability.

Additionally, one of the distinctions between this research and others is the list of indicators used for analysis. The research indicators have been chosen in accordance with the Iranian context and considering data availability. Differences in the indicators lead to variations in the outcomes. Inadequate data access, as well as the complexity associated with calculating social vulnerability indicators, are among the challenges confronting social vulnerability studies in Iran [79]. COVID-19 first spread among groups with a higher economic and social status in the case of Iran, especially in Khuzestan Province. Thus, the neighborhoods and urban areas with higher economic and social status reported a higher number of COVID-19 patients [115,116]. Previous studies have also reported a higher rate of Covid-19 among people with middle and high income in richer countries with higher interpersonal interactionsat work and more frequent trips inside and outside the country [117,118].

The major cities and provincial capitals were the first to be hit with COVID-19 in Iran. It then gradually spread to surrounding cities and villages. Accordinglyection rates were initially high in the capital of Khuzestan province but quickly grew in all villages and cities within a few months.

A review of the experiences and the available data indicate that the marginal and deprived groups, especially in Khuzestan as a multi-hazard province, and those who had less economic and social power have suffered the most damage from natural disasters such as floods and earthquakes. Although this research indicates that social vulnerability does not result in increased COVID-19 rates, the COVID-19 outbreak has impacted vulnerable areas, including suburban neighborhoods with informal jobs, residents of worn-out neighborhoods, and villagers whose livelihoods are reliant on urban areas.

It has been also well-recognized that vulnerable people have fewer opportunities to work remotely [119] and their jobs mostly require physical presence. According to Das et al. [60]; residents in slums and informal settlements worldwide are the most vulnerable to COVID-19. In general, the social vulnerability of disadvantaged groups prevents them from complying with social distancing regulations [73] because these groups do not have enough income and savings and are forced to go to work or work on the streets to sustain their livelihood. Previous studies have shown that income is a key factor in compliance with social distancing measures [120].

The COVID-19 outbreak has created many challenges for people in poor areas and vulnerable groups in Iran and Khuzestan Province. Marginal areas and urban slums sometimes face many environmental health problems due to distance from urban services or poor urban infrastructure. Due to the lack of sanitary items such as masks, gloves, and disinfectants and the high price of these items, the poor, the needy, and disadvantaged social groups cannot access them easily. Socially vulnerable groups such as street vendors, kids involved in child labor, beggars, garbage collectors, drug addicts, and homeless people are more likely to be exposed to the virus and spread it among other community members. Due to the high costs of medical services and insurance coverage problems, the deprived and needy classes usually have fewer visits to the doctor than the rich. It makes them more vulnerable, especially against COVID-19.

6. Conclusion

This research is one of the first in Iran to examine links between social vulnerability and COVID-19. The social vulnerability was measured quantitatively in 27 counties of Khuzestan province, and then its relationship with the COVID-19 rate was explored. The findings indicate that counties' social vulnerability varies little according to their rural and suburban population densities, and counties with a high total population density and a low rural population density are more vulnerable. There is a non-significant negative correlation between the level of social vulnerability in counties and the COVID-19 rate.

The findings demonstrate that COVID-19 is a complex issue that varies according to various factors, including management and planning policies, social behaviors, and participation in coping programs. Other areas of vulnerability can be assessed by utilizing decision-making models and indicators, particularly residential indicators. The research findings can also be applied to administration and policy-making, providing managers with an insights on the impact of residents' socio-economic conditions and COVID-19 rates. The rate of COVID-19 cases in the province is unrelated to the integrated social vulnerability index, and all counties must adhere to protocols and health regulations. Given that quarantine conditions and mobility restrictions have been implemented in all counties, those with a higher degree of social vulnerability require more attention. The province's authorities should provide special support packages to such groups. Among the populations targeted by support programs are families covered by institutions, working children, drug addicts, street sweepers, the homeless, female-headed households, informal workers, and day laborers. According to the research findings, specific support measures should be implemented for counties with high rates of rural population, such as Andika and Izeh, as well as counties with high rates of informal settlements, such as Ahvaz, Mahshahr, Abadan, and Khorramshahr. In the counties where the number and distribution of settlements are high, and access is difficult, the provision of sanitary items and medical equipment related to COVID-19, and the establishment of temporary health monitoring centers are among the measures that are needed to control new waves of the pandemic.

Since this study adopted a spatial and location-based perspective and addressed challenges faced by deprived and marginal areas, the following preventive and support measures could be taken into account:

• -Adopting appropriate decision-making process for the accurate identification of counties, slums, and the outskirts that need to be prioritized in efforts aimed at controlling the COVID-19 infection,
• -Making effective context-sensitive decisions to provide support and health services (e.g., provision of disinfection equipment, masks, gloves, and disinfectants, payment of subsistence aid, and distribution of health items and health aid) for families living in poor neighborhoods and on the outskirts of counties,
• -Increasing spatial justice and expanding the coverage of urban services to residential areas with less access to municipal services.
• -Targeted subsidy payments to low-income groups, especially female heads of households and people who are not covered by social and medical insurance.

The findings of this study can contribute to the growing knowledge of social vulnerability. The study used objective indicators and existing data to measure social vulnerability and the COVID-19 incidence rate. Using objective data results in fewer errors than self-reported and subjective data. The present study also measured social vulnerability as a composite index and addressed it as a variable. It also addressed residential distribution indicators (population distribution) as one of the indicators for measuring social vulnerability because the spread of settlements and population makes it difficult to access services, especially in rural areas. Moreover, like other related research projects, this study reflected social inequalities and deprivations that existed before the COVID-19 crisis [69].

It should be acknowledged that this research had some limitations. First, it was impossible to evaluate the relationship between social vulnerability and the COVID-19 rate across all of the province's cities and rural areas due to the difficulty of acquiring data on social vulnerability per city and village. Such detailed analysis could have provided a better analysis of the vulnerability conditions. Second, not all urban and rural areas have equal access to COVID-19 test laboratories. Accordingly, the number of reported cases in some areas may be less than the real numbers. Third, and related to the previous point, COVID-19 test reports contain some statistical inaccuracies. Fourth, it was not possible to access the land use layers for a detailed examination of the level of access to services, especially healthcare services. Fifth, it was not possible to separate the data into urban and rural parts and measure the relationship between the research variables for rural and urban areas separately. Finally, access to information at the individual level was not possible. The condition of infected individuals is a better indicator of the relationship between socio-economic status and COVID-19 cases.

It is recommended that future studies examine the relationship between social vulnerability and COVID-19 mortality rates at more granular levels. Social vulnerability should be assessed using variables such as the prevalence of informal employment, the population of informal settlements, and the rate of car ownership. To obtain more accurate measurements, the COVID-19 rate should be investigated separately for urban and rural populations and based on slum and suburb locations. The socio-economic aspects and access levels are evaluated more favorably at smaller scales [113].

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.